Re: yipeee!

joel-garry_at_home.com (Joel Garry) wrote in message news:<91884734.0402061511.11374f42_at_posting.google.com>...
> I guess there's no such thing as a proper enterprise environment. One
> of my favorite personal war stories is the electrician who reached in
> and turned off the UPS for Superdome, numerous HP's, etc.

I have also seen something similar. However, these are exceptions, that happens maybe once in three years or something like that. 99.99(9)% of time everything is running smoothly, so benchmarking state that occurs in 0.0(0)1% of time is not probably very justificated.

> Benchmarks are ok since they state their parameters. We all know how
> to tell when a marketing person is lying...

The first thing that I want to know when I see some new benchmark is who paid for it. In complex IT environment there is a lot of variables that can be changed according to needs of tested product. You can see a lot of benchmarks that favours some subset of operations because the benchmarked product is better while dealing with them than its competitors. Microsoft is quite a good example of this. If we're talking about TPCs then my biggest problem with them is that they do not check if the final solution is feasible from production point of view (Microsoft's mega clusters) or is just support and operating nightmare.
Also could be quite funny if you take a numbers from any price/performance benchmark and go to the nearest vendor and want used HW for the price that was used for calculation of price/performance measurement. :-)

--
Dusan Bolek